The demand for up-to-date cartographic information is continuously increasing. A major factor for this development is new technology such as the internet, broadband, GPS and 3G telephony. These technologies make it possible to offer new services based on digital maps. To satisfy this large marked, the great challenge ahead is to offer up-to-date cartographic databases. The information in today’s databases has been mainly collected from existing analog maps and aerial photos. Digital images from both planes and satellites will take over and be used for making new data as well as updating existing databases. The efficiency of the existing computer systems for converting the images to GIS data is low. This is due to the need for human interpretations and digitizing. Hence, much interactive time is spent in front of a workstation. This makes the production expensive and slow and consequently desired information may be left out.

The ImCap project develops new methodology for semiautomatic data capturing in high-resolution image data. The new approach will make it possible to more frequently update the GIS databases as well as capture a larger amount of data more reliable than with existing systems.

The research in the projects spans over many areas from structural detections to object modeling to machine learning. Our semi automatic approach to optimally utilize the strength of humans and computers is a challenge and innovative in itself. The semi-automatic approach sets certain demands to both theory and methods to be developed. A number of research challenges are included such as the development of flexible object models, estimation of the probability of an object model, interactive model adoption and adoptive algorithms learning from the operator.